Second devices and screen sharing: closing the quiet cheating channels
Why the second screen beats the lockdown
A lockdown browser controls software on one machine. It cannot see a phone propped against a monitor stand, a smartwatch notification, or an earpiece. Screen sharing is the more serious channel because it scales the help: one strong candidate can coach several others in real time, reading questions as they appear and feeding answers back through a channel the exam never sees.
Remote desktop tools blur the line further. Some are ordinary IT support software, which makes blanket blocking a support headache, and determined candidates can run the exam inside environments designed to look clean. The honest conclusion is that device control alone has a ceiling, and exam design has to carry the rest.
Room and behaviour signals that help
Webcam room scans before the exam establish what is in reach: extra screens, notes, other people. Random short scans during the exam, announced in the rules in advance, raise the risk of using a device openly. Gaze and behaviour flags add weak signals; a candidate repeatedly reading below camera line in a steady rhythm deserves a review flag, not a verdict.
Audio matters too. A helper in the room or on a call often leaks through the microphone as whispers, keyboard sounds that do not match the candidate's hands, or answers arriving in suspicious pauses. Reviewers listening to flagged segments catch arrangements that video alone misses.
Exam design is the strongest control
Questions that assume open resources change the game. If the exam rewards applying knowledge to a fresh scenario, a helper relaying looked up facts adds little speed. Timed sections with question pools, so neighbours and coached groups hold different papers, break the economics of one helper serving many candidates at once.
Short oral follow ups for high stakes results are the audit that scales worst for cheaters. Asking a candidate to explain two of their own answers live, for five minutes, is nearly impossible to pass on borrowed work and entirely fair to honest candidates, who experience it as a conversation about their own thinking.
Setting rules candidates can actually follow
Rules should name the reality plainly: one screen, no phone in reach, no earpieces, no one else in the room, and software that shares the screen is prohibited. Vague honesty pledges leave honest candidates unsure whether a phone face down on the desk is an offence, and unsure candidates either over comply anxiously or under comply casually.
Institutions should also decide the response ladder before exams run: which flags pause an exam, which trigger review afterwards, and what evidence a finding needs. Consistent, pre published responses are what make enforcement defensible when a result is challenged.
Keep reading: Impersonation in remote exams: identity checks that actually hold up, How long should remote exam recordings and proctoring data be kept?, and How should extra time and breaks work in proctored online exams? cover the nearby cases in this series.
Frequently asked questions
Can a lockdown browser detect a second device?
Not directly. It controls the exam machine only. Detection comes from webcam room scans, behaviour and audio signals, and exam design. That is why second device rules and random checks during the exam matter as much as the browser itself.
Why is screen sharing such a serious channel?
Because it exports the whole exam to a helper anywhere in the world, in real time. One knowledgeable person can coach multiple candidates simultaneously, and the help never appears on the exam device the institution is monitoring.
What exam design changes reduce second device value?
Fresh scenario based questions, per candidate question pools, tight section timing, and short oral follow ups on submitted answers. These reduce what a helper can add and make borrowed performance easy to expose.
Should gaze tracking flags fail a candidate automatically?
No. Gaze patterns are weak evidence affected by thinking habits, accessibility needs, and camera position. They justify human review of the session recording, and findings should rest on the full evidence, never on a single automated flag.